Potent DWI connectome markers of verbal and non-verbal functions in children with unilateral lesions
Poster No:
1543
Submission Type:
Abstract Submission
Authors:
Min-Hee Lee1, Jeong-Won Jeong1, Nore Gjolaj1, Hiroshi Uda1, Michael Behen1, Aimee Luat1, Eishi Asano1, Csaba Juhasz1
Institutions:
1Wayne State University, Detroit, MI
First Author:
Co-Author(s):
Introduction:
Unilateral brain lesions in children longitudinally affect neural plasticity mechanisms of neurocognitive functions including verbal IQ (VIQ) and non-verbal IQ (NVIQ).[1-3] This study aims to develop an imaging marker that can accurately differentiate two DWI connectomes (DWICs) related to VIQ and NVIQ from children with unilateral brain lesions, such as lesional epilepsy and Sturge-Weber syndrome (SWS; a venous vascular disorder), by using a deep learning tract classification approach with clinically acquired DWI tractography.
Methods:
Three cohorts: 6 healthy children with typical development (TD, 12.3±3.8 years old, 3 boys), 52 children with epilepsy (11.7±3.5 years old, 25 boys; 21/13/18 left/right/non-lesional epilepsy), and 11 children with SWS (13.3±2.7 years old, 5 boys; 3/8 left/right-hemisphere involved SWS) were recruited and completed 3T DWI tractography and neuropsychological language assessments including VIQ and NVIQ tests. 3T DWI tractography scans were also collected from 29 TD (11.6±3.3 years old, 14 boys) as baseline data. To ensure reproducibility of DWIC, our previous DCNN-based tract classification[4], which effectively removes false-positive tracts (e.g., wiggly shaped tracts), was extended to the iFOD2-ACT whole-brain tractography[5,6] of individual subjects. This approach yields a whole-brain backbone DWIC, Si,j, in which each element defines an average factional anisotropy (FA) value of true-positive tract classes, Ck=1,2,…,1477, consisting of a series of tracts connecting i and j regions in the AAL atlas. Briefly, for each tract of the training Ck, our DCNN model was designed to learn 3-D coordinates of 100 equal-number tract segments by minimizing focal and center loss. Each VIQ and NVIQ network comprises specific axonal connectivity edges derived from Si,j of 69 subjects (i.e., 6 TD, 52 epilepsy patients, and 11 SWS patients), with elements of Si,j that significantly correlate with VIQ and NVIQ scores across subjects (p-value of correlation coefficient<0.001 after controlling for age and sex). Global efficiency (GE)[7], which measures efficiency of parallel information flow in a network, was calculated for each VIQ and NVIQ intra-hemispheric network. To assess how far the GE marker value deviates from baseline, Z-scores of GE values were calculated. A non-parametric ANOVA test was applied after controlling for age and sex to assess the difference in Z-scores of GE values between ipsilateral and contralateral intra-hemispheric network. To explore the potential of this imaging marker to assess neurocognitive outcomes, the strength of the linear relationship between the Z-scores of GE values and IQ scores was evaluated by Pearson's correlation.
Results:
We could successfully differentiate the VIQ network with 24 key nodes and the NVIQ network with 22 key nodes (Fig. 1). Figure 2A presents that Z-scores of GE in the ipsilateral IQ network were lower than those in contralateral IQ network. Specifically, the Z-score of GE in right VIQ and NVIQ networks in right epilepsy (n=21) was significantly lower than that of left epilepsy (n=13) (p=0.04 and p=0.03). The VIQ score was significantly correlated with the Z-score of GE in the left VIQ network (r=0.36, p=0.03) across all subjects (Fig. 2B). Additionally, the NVIQ score was significantly correlated with the Z-score of GE in the right NVIQ network (r=0.37, p=0.05) across all subjects (Fig. 2B), while there was no significant correlation between VIQ/NVIQ scores and Z-scores of GE for separate cohorts (Fig. 2C).
Conclusions:
We provided preliminary evidence that lower efficiency of neural information flow in the ipsilateral VIQ and NVIQ networks can be used as potential markers for assessing lower verbal and non-verbal functions via effectively differentiating VIQ and NVIQ networks disrupted by the ipsilateral lesions in children with unilateral lesions. The reproducibility of this finding should be further evaluated in a larger cohort.
Disorders of the Nervous System:
Neurodevelopmental/ Early Life (eg. ADHD, autism) 2
Higher Cognitive Functions:
Reasoning and Problem Solving
Language:
Reading and Writing
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 1
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
White Matter Anatomy, Fiber Pathways and Connectivity
Keywords:
Cognition
Epilepsy
Language
Pediatric Disorders
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
Other - Sturge-Weber syndrome
1|2Indicates the priority used for review
Provide references using author date format
[2] Behen, M.E. (2011), ‘Brain damage and IQ in unilateral Sturge-Weber Syndrome: Support for a “fresh start” hypothesis’, Epilepsy & Behavior, vol. 22, no. 2, pp. 352-357.
[3] Chugani, H.T. (1996), ‘Functional brain reorganization in children’, Brain and Development, vol. 18, no. 5, pp.347-356.
[4] Lee, M.H. (2019), ‘Improving reproducibility of diffusion connectome analysis using deep convolutional neural network model’, Proceedings of the International Society for Magnetic Resonance in Medicine, vol. 27, pp. 3578.
[5] Tournier, J.D. (2010) Improved probabilistic streamlines tractography by 2nd order integration over fibre orientation distributions’, Proceedings of the International Society for Magnetic Resonance in Medicine, vol. 18, pp. 1670.
[6] Smith, R.E. (2012), ‘Anatomically-constrained tractography: Improved diffusion MRI streamlines tractography through effective use of anatomical information’, NeuroImage, vol. 62, no. 3, pp. 1924-1938
[7] Latora, V. (2001), ‘Efficient behavior of small-world networks’, Physical Review Letters, vol. 85, no. 19, pp. 198701.